In most organisations, the decisions that matter most are made well before the evidence meant to inform them actually arrives. A leader commits capital to a new market, repositions a product, or moves on a competitive opening, while the study that should have guided the choice arrives weeks later. It is rigorous and well-built, but too late to change a call that has already been taken.
This is a central problem facing businesses in Singapore. Decisions now move faster than the research designed to support them. The familiar cycle, where a question is scoped, a study fielded, the data cleaned, and an answer delivered within three to six weeks, no longer matches the speed at which markets shift and customers change. Bets on pricing, positioning, market entry, and customer experience are made on instinct because waiting for the evidence means missing the window to act on it.
The way this gap gets closed is now being rebuilt. Global Qualtrics data shows that 95% of market research professionals already use AI for tasks such as drafting questions, summarising responses, or cleaning data. The larger shift is just ahead. Within the next six to twelve months, adoption of agentic research systems, where AI runs the full research lifecycle instead of single tasks, is set to nearly triple, from 15% to 44%. The use of synthetic data is projected to climb from 41% to 62%.
Seventy-two per cent of researchers anticipate that agentic AI will assume full responsibility for critical research phases, including study design, fielding management, and analytical planning.
- End-to-End Execution: 70% of respondents believe Research Agents will facilitate more than half of all end-to-end research projects in the near future.
- Early Adoption: Significantly, 75% of researchers report they are already actively experimenting with or regularly integrating Agentic AI within their current methodologies.
Singapore organisations are on the same path. Most have already put AI to work on the narrow tasks, generating questions, summarising responses, processing data. What is changing now is the scope of what AI handles: moving from assisting with parts of a study to running it, from the opening plan through to the finished, synthesised insight. That capability can move quickly, from something teams experiment with to something they rely on.
Where it bites
See also: China pushes all school levels to teach AI in Xi’s tech drive
For the leaders who own commercial decisions, the friction shows up in three ways.
The first is speed. A decision on whether to enter a segment, reprice a product, or back a new proposition cannot wait three weeks when the call has to be made next week. When insight arrives too slowly, the decision simply gets made without it.
The second is capacity. Many Singapore-based teams support regional or global decisions with a lean headcount. The volume of questions has grown faster than the people available to answer them. Functions from marketing to strategy compete for the same limited research bandwidth, and AI that automates only fragments of the work leaves the core problem untouched.
See also: OpenAI limits release of new model under pressure from US
The third is trust in the output. As AI fills organisations with faster answers, a new risk appears: output that sounds authoritative without being grounded in real data. That is more dangerous than slow insight, because it reaches the decision-maker and gets acted on before anyone thinks to check it.
The opportunity, not just the fix
Done well, this reshapes the operating model itself. A business question, posed by a leader, triggers an agent that designs the study, fields it across both human and AI-modelled synthetic audiences, draws on everything the organisation has already learned, and returns a decision-ready answer in hours. Methods that once required specialist consultants, such as advanced segmentation or trade-off analysis, have become part of the standard workflow. The teams closest to the customer take on a more central role, shaping the research as it runs and turning its findings into recommendations the business can act on.
This is where Singapore's conversation around AI is usefully maturing. The interest is moving past novelty toward AI that produces outcomes worth acting on. For a business leader, that is the only test that matters. Speed means nothing if the answer does not hold up.
The integration of synthetic data generation can extend the return on investment even further, giving researchers access to rapid insights, enabling them to follow up with more targeted studies, and delivering richer insights in real-time and more cost-effectively than traditional methods.
One use case is concept testing. Brands are increasingly challenged to replace guesswork with data-driven precision while operating within lean research budgets. Industry leaders like Fonterra are meeting this challenge by integrating Qualtrics’ Synthetic and Conversational AI. The combination of these AI technologies helps deliver superior data quality and real-time, personalised follow-ups to uncover the 'why' behind consumer preferences. By automating this depth of insight, brands can achieve a holistic understanding of market potential while significantly reducing the costs associated with traditional, labour-intensive research.
The early evidence is encouraging. Booking.com used synthetic panels to run a psychographic segmentation study and found the same spread of variance within segments as a human sample, validated through parallel studies, at a speed a human-only approach could not match. The result sharpened how the business understood and reached its customers.
Telling substance from noise
To stay ahead of the latest tech trends, click here for DigitalEdge Section
For leaders weighing where to invest, a handful of standards separate a genuine research agent from task automation in better packaging.
A strong system embeds proven methodology, so when more people across the business run research, they still get expert-grade studies and quality holds as use widens. A capable system knows what it does not know. When past research cannot answer a question, a reliable agent says so and commissions new fieldwork, ensuring that every answer rests on real data.
The synthetic audience has to be built on real human data. Panels modelled on validated survey responses behave like real populations and reach the same conclusions a human sample would. That grounding is what makes a synthetic panel research-grade.
The work should carry through to action. The value of research lies in the decision it drives, so a capable agent takes a finding through to a clear recommendation and closes the gap between what the research shows and the call it should inform.
Why now
The pieces have converged and decision cycles have shortened to the point where slow insight carries a real competitive cost. The technology can now run the full workflow, not only parts of it. And across Singapore’s economy, the expectation is shifting steadily toward decisions grounded in data and made at speed.
The organisations that build this capability over the next year will gain more than speed. They put evidence in front of leaders at the moment of decision, when it can still shape the call. Over time, that reshapes how the business decides, putting what the customer and the market are actually saying at the front of every choice. In this model, the expertise inside these teams scales. It is brought to the centre of every decision, present at the moment the call is made.
Hui Ching Tan is the head of Research Insights APJ at Qualtrics
